1. I'm on beaTunes 5.1.5. I just did this 2 days ago, so I'd say it's still an issue. :) I'll work on narrowing down which update has what effect on the files. When I did the update, I was letting the analyzer perform several different updates.2. 99% of my music library is mp3. These files are mp3.
I specified a new library with 1 album in it. I analyzed each of these items separately for 1 song: BPM, key, color, mood, danceability, fingerprint, in that order. In other words, I selected 1 song and ran the analysis six times, selecting a different item each time. In every case, the modified timestamp was updated but not the file created timestamp (which is good).
Then I ran Analyze All Songs on the same library, specifying to not overwrite values, again running the analysis for just one aspect at a time. Remember that 1 song had already been analyzed. Again, each time, the Modified timestamp was updated and the Created timestamp was left alone (still good).
Then I specified another new library folder with 1 album in it. This time I ran Analyze All Songs on the same 6 parameters as before. This time the Modified AND Created timestamps were updated. Not good. So, maybe if an analysis has been run before, the Created timestamp doesn't change, but on a "virgin" library, it does?
I hope this helps you to continue to evaluate this situation.
hendrik on 16 May, 2018 02:31 PM
just FYI: id3 v2 tagging works by embedding tags at the beginning of an audio file. Since there isn't always enough space there to add meta information without overwriting audio information, sometimes space has to be made by making the file longer and shifting audio information towards the back. In this process JAudioTagger used to create a new file, because it's simpler.
I just did some digging in the JAudioTagger code and adjustPadding() apparently still creates new files. I'll try to get this fixed. I wasn't aware of it.
I think the biggest "news" that I found was that in one case, new files were not created, and in a different case they were... for essentially the same type of update. The only difference that jumped out at me was in one case, when I selected Analyze All Files, one file has already been analyzed and updated. In the other case, no files had been previously analyzed. Granted, they were different file sets, so there might be something significantly different about the two sets of files themselves. I'll look into that.
Thank you very much, Hendrik. I repeated the analysis on the same file set and am happy to report that the file identities ("create date") were unchanged. I will continue to play with the release as time permits, but the initial result looks very good. Thank you for addressing this so quickly.